DE102014204876A1 - Inspection procedure and inspection device - Google Patents

Abstract

An inspection method and apparatus, which comprise virtually dividing a sample in which a plurality of chip patterns are formed into a plurality of strip-shaped strips along a predetermined direction to detect an optical image of the chip pattern in each of the strips filtering based on design data of the chip pattern to produce a reference image corresponding to the optical image, comparing the chip pattern using a die-to-database method, and comparing a repetitive pattern area in the chip pattern using a cell method Determining a dimensional difference or / and a dimensional ratio between a pattern of the optical image and a pattern of the reference image compared with the pattern of the optical image by the to-database method; and determining a dimensional distribution of the plurality of chip patterns from the dimensional difference or / and the dimension ratio.

Description

CROSS-REFERENCE TO RELATED APPLICATION

The entire revelation of Japanese Patent Application No. 2013-055500 , filed on Mar. 18, 2013, including specification, claims, drawings and abstract on which the priority is based upon the cooperation agreement of the present application is hereby incorporated in its entirety.

FIELD OF THE INVENTION

The present invention relates to an inspection method and an inspection apparatus.

BACKGROUND

With high integration and high capacity of Large Scale Integration (LSI), a circuit scale required for a semiconductor device is increasingly narrowed. For example, in recent typical logic devices, it is necessary to form a pattern with a line width of several tens of nanometers.

It is necessary to improve a production yield of the expensive LSI in a production process. In the semiconductor element, during the production process, a source design pattern (ie, a reticle, hereinafter collectively referred to as a mask) in which a circuit pattern is formed is called by a reduction projection exposure apparatus called a stepper or a scanner is exposed and transferred to a wafer. A shape defect of a mask pattern can be listed as a large factor, which reduces a production yield of the semiconductor element.

The finer the dimension of an LSI pattern formed on the wafer, the smaller the shape defect on the mask pattern becomes. With absorption of fluctuations of various process conditions by improving the dimensional accuracy of the mask, it is necessary to detect a defect of extremely small pattern in a mask inspection. At this point, it is also necessary to determine the defect taking into consideration fluctuation in line width dimensions or position shift amount of the pattern in a mask surface. For example, this discloses Japanese Patent No. 4236825 an inspection device that can detect a fine defect in the mask.

Examples of defect detection techniques include a die-to-database comparison method and a die-to-the-compare method. In the die-to-database comparison method, a reference image generated from design pattern data used in the mask production and an optical image of the actual pattern on the mask are compared with each other. In the die-to-die comparison method, in the case that a plurality of chips having the identical pattern configuration are arranged in one part or in the whole in the identical mask, the optical images having the identical pattern in chips of the different masks are compared with each other.

A cell comparison method may also be cited as another defect detection technique. The cell comparison method is effectively used in the case where a repetitive pattern called a cell exists in the mask. In the die-to-compare method, the chips each formed in the mask are compared with each other. On the other hand, in the cell comparison method, the repetitive patterns such as memory mats, namely the cells, are compared with each other in a chip. For example, the defect is inspected by the cell comparison method in a memory cell group of a DRAM (Dynamic Random Access Memory) element in which the repetitive pattern is formed. On the other hand, a logic element in which the repetitive pattern does not exist is inspected by the to-the-match method, in which the pattern of the logic element is compared with the pattern of a dummy logic element in an inspection dummy pattern that is on a predetermined position is provided in the mask. Nowadays, as the demand for logic-embedded memory increases, sometimes both the die-to-compare method and the cell compare method are performed in a concurrent inspection process (see, for example, US Pat Japanese Patent No. 444768 ).

Conventional mask inspection is aimed at detecting a shape defect of the pattern, and a defect determination algorithm suitable for detecting the shape defect of the pattern, and a defect recording method are prepared. In the mask inspection apparatus, a function of detecting the defect caused by the fluctuation in the line width of the pattern is improved to meet the challenge of a lack of LSI production margin caused by the fluctuation in the line width. However, in a current mask pattern, the shape defect or the dimension of the defect found as the cause of fluctuation in the line width becomes substantially equal to the fluctuation of the line width (line width distribution) on the entire surface of the mask. Therefore, the number of detected defects becomes large.

In a process of generating the reference image in the die-to-database comparison method, the filtering of the optical image of a typical pattern position in the mask, namely the learning process of a filter coefficient on the design pattern data, is performed, whereby the reference image becomes the pattern image that has a Line Width Tendency mimicking the pattern line width of a region where the learning process is performed. Therefore, the line width dimension has a distribution in the mask even in the to-database comparison method, and the optical image and the reference image are compared with each other with a linewidth influence (deviation) of the pattern in inspecting the region, which differs a pattern linewidth of the pattern line width of the region where the learning process is performed. As a result, the shape defect to be detected or the fluctuation in the line width can not be detected, or a shape and line width that does not require detection is detected as a defect.

In addition, in the die-to-the-comparison method, the patterns having the linewidth variation (deviation) are compared with each other when the chips in the regions having the different line widths are compared with each other. Accordingly, the defect to be detected can not be detected or detected shape and line width, which do not require detection, as a defect.

An object of the invention is to provide an inspection method and apparatus that are capable of reducing the detection of an unnecessary defect because the defect to be detected is detected.

Other challenges and advantages of the present invention will become apparent from the following description.

SUMMARY OF THE INVENTION

According to one aspect of the present invention, an inspection method comprising virtually dividing a sample in which a plurality of chip patterns are formed into a plurality of strip-shaped strips along a predetermined direction to form an optical image of the chip pattern in each of the strips performing filtering based on design data of the chip pattern to produce a reference image corresponding to the optical image, comparing the chip pattern using a die-to-database comparison method, and comparing a repetitive pattern portion in the chip pattern using a cell method, determining a dimensional difference or / and a dimensional ratio between a pattern of the optical image and a pattern of the reference image compared with the pattern of the optical image by the to-database comparison method, and determining a dimension distribution of the plurality from chip patterns from the Dimensional difference or / and the dimensional ratio, wherein with respect to a position determined by the comparison of the die-to-database comparison method as a defect, a result of the die-to-database comparison method is stored when a dimension distribution from the position to a previous position where the dimensional difference or / and the dimension ratio is detected falls within a predetermined range, by comparing the dimension distribution with a dimension distribution in the strip containing the defect-determined position or a dimension distribution of the chip pattern, the is assumed to be the strip in which the dimensional difference is determined in advance of the strip concerned, and a result of the cell method is stored instead of the result of the die-to-database comparison method, if the dimension distribution from the position determined as a defect to the previous one Position where the dimension difference and / or the dimensional ratio is determined to exceed the predetermined range by comparing the dimension distribution with the dimension distribution in the strip containing the position determined as a defect, or the dimension distribution of the chip pattern adopted from the strip in which the Dimension difference is determined in advance of the affected strip.

Further, with respect to this aspect of the present invention, an inspection method wherein the result of the die-to-database comparison method is stored to the previous position regardless of the dimension distribution from the position determined as a defect, wherein the dimension difference or / and the dimension ratio are obtained if the result of the cell comparison does not exist because the repetitive pattern part does not exist at the position determined as a defect by the comparison of the die-to-database comparison method.

Further, according to this aspect of the present invention, an inspection method wherein the dimensional difference is a difference in the line width between the pattern of the optical image and the pattern of the reference image or a difference of a distance between the patterns of the optical image and a distance between the patterns of the reference image ,

Further, according to this aspect of the present invention, an inspection method wherein the dimensional ratio is one Line width ratio of the pattern of the optical image and the pattern of the reference image is or a ratio of a distance between the patterns of the optical image and a distance between the patterns of the reference image.

In another aspect of the present invention, an inspection method comprising determining an optical image of a sample having a plurality of chip patterns formed thereon, performing filters based on design data of the chip pattern, around a reference image corresponding to the optical image comparing the chip pattern by a die-to-database method and comparing a repetitive pattern part in the chip pattern by a cell method, determining a dimensional difference or / and a dimension ratio between a pattern of the optical image and a pattern of the Pattern of the optical image by the reference image compared to the database method, and determining a dimension distribution of the plurality of chip patterns from the dimensional difference or / and the dimensional ratio, with respect to a by comparing the die-to-database method as Defect specific position, a result of the die-to-database v method, when comparing the dimensional distribution with a dimension distribution in a chip or a dimension distribution between chips, the dimension distribution from the position to a previous position at which the dimensional difference or / and the dimension ratio is detected falls within a predetermined range, and a Result of the cell method is stored instead of the result of the die-to-database method, when the dimension distribution from the position determined to be a defect to the previous position where the dimension difference and / or the dimension ratio is determined exceeds the predetermined range, by comparing the dimensional distribution with the dimensional distribution in the chip or the dimension distribution between the chips.

Further, to this aspect of the present invention, an inspection method wherein the result of the die-to-database method is stored at the previous position regardless of the dimension distribution from the position determined as a defect, wherein the dimension difference or / and the dimension ratio is determined if the result of the cell comparison does not exist because the repetitive pattern part does not exist at the position determined as a defect by comparing the die-to-database method.

Further, to this aspect of the present invention, an inspection method wherein the dimensional difference is a difference in the line width between the pattern of the optical image and the pattern of the reference image or a difference of a distance between the patterns of the optical image and a distance between the patterns of the reference image.

Further, to this aspect of the present invention, an inspection method wherein the aspect ratio is a line width ratio of the pattern of the optical image and the pattern of the reference image, or a ratio of a distance between the patterns of the optical image and a distance between the patterns of the reference image.

In another aspect of the present invention, an inspection method comprising virtually dividing a sample in which a plurality of chip patterns are formed into a plurality of strip-shaped strips along a predetermined direction to obtain an optical image of the chip pattern in each of the strips; Comparing the chip pattern by a die-to-die method and comparing a repetitive pattern part in the chip pattern by a cell method, determining a dimensional difference or / and a dimension ratio between a pattern of the optical image and a pattern of the reference image, as compared with FIG Pattern of the optical image by the die-to-die method, and determining a dimensional distribution of the plurality of chip patterns from the dimensional difference or / and the dimension ratio, wherein with respect to a position, which by comparing the die-to-die Method is determined to be a defect, a result of the die-to-die method ns is stored when a dimension distribution from the position to a previous position at which the dimensional difference or / and the dimension ratio is detected falls within a predetermined range by comparing the dimensional distribution with a dimension distribution in the strip containing the position determined as a defect , or a dimension distribution of the chip pattern adopted from the strip in which the dimensional difference is determined in advance of the strip concerned, and a result of the cell method is stored instead of the result of the to-the-the-method if the dimension distribution is off the position determined as a defect to the previous position where the dimensional difference or / and the dimension ratio is determined exceeds the predetermined range, by comparing the dimension distribution with the dimension distribution in the strip containing the position determined as the defect , o the dimension distribution of the chip pattern adopted from the strip, the dimensional difference being determined in advance of the strip concerned.

Further to this aspect of the present invention, an inspection method wherein the result of the die-to-die process is stored regardless of the dimension distribution from the position determined as a defect to the previous position where the dimension difference or / and the dimension ratio is determined is when the result of the cell method does not exist because the repetitive pattern part does not exist in the position determined by the comparison of the die-to-die method as a defect.

Further, to this aspect of the present invention, an inspection method wherein the dimensional difference is a difference in the line width between the patterns of the optical images or a difference in the distance between the patterns of the optical images.

Further, to this aspect of the present invention, an inspection method wherein the aspect ratio is a line width ratio of the patterns of the optical images or a ratio of a distance between the patterns of the optical image.

In another aspect of the present invention, an inspection apparatus includes an optical image detection unit (A) that virtually divides a sample into a plurality of stripe-shaped stripes along a predetermined direction to detect an optical image of the sample in each of the stripes, a reference image generation unit ( 112 ) performing filtering based on design data of the chip pattern formed on the sample to produce a reference image corresponding to the optical image, a first comparator (Fig. 108a ) which displays the chip pattern of the optical image output from the optical image detecting unit (A) with the chip pattern of the reference image generating unit (Fig. 112 ) compares the reference image output by a database-to-database method, a second comparator ( 108b ) comparing repetitive pattern parts in the chip pattern of the optical image output from the optical image detecting unit (A) using a cell method, a dimension difference / dimension ratio detecting unit (FIG. 125 ), which determines a dimensional difference or / and a dimensional ratio between a pattern of the optical image and a pattern of the reference image, as compared with the pattern of the optical image by the die-to-database method, a dimension distribution detecting unit (FIG. 126 ) which determines a dimension distribution of the plurality of chip patterns from the dimensional difference or / and the dimensional ratio, which is calculated from the dimensional difference / dimension ratio detection unit (FIG. 125 ) and a controller ( 110 ), which is a result of the first comparator ( 108a ) when, in relation to a location determined by the comparison of the first comparator ( 108a ) is determined as a defect, a dimension distribution from the place to a previous place, the dimensional difference or / and the Dimension ratio by the dimension difference / dimension ratio detection unit (FIG. 125 ), falls within a predetermined range, by comparing the dimension distribution with a dimension distribution in the strip containing the space determined as a defect or a dimension distribution of the chip pattern suspected from the strip in which the dimension difference is in advance affected strip, and a result of the second comparator ( 108b ) instead of the result of the first comparator ( 108a ) stores, when the dimension distribution of the space designated as a defect to the previous place where the dimensional difference or / and the dimensional ratio by the dimension difference / dimension ratio detecting unit (FIG. 125 ) exceeds the predetermined range, by comparing the dimensional distribution with the dimension distribution in the strip containing the defect-specific space or the dimension ratio of the chip pattern suspected from the strip in which the dimension difference is detected in advance of the stripe concerned.

Further, to this aspect of the present invention, an inspection apparatus comprising: an optical image detecting unit (A) that virtually divides a sample into a plurality of stripe-shaped stripes along a predetermined direction to detect an optical image of the sample in each of the stripes; Comparator ( 108a ) comparing the chip patterns of the optical image output from the optical image detecting unit (A) by a die-to-die method, a second comparator ( 108b ) comparing repetitive pattern parts in the chip pattern of the optical image output from the optical image detecting unit (A) by a cell method, a dimension difference / dimension ratio detecting unit (FIG. 125 ), which determines a dimensional difference or / and a dimensional ratio between a pattern of the optical image and a pattern of the reference image compared to the pattern of the optical image by the die-to-die method, a dimension distribution detecting unit (FIG. 126 ) which determines a dimension distribution of the plurality of chip patterns from the dimensional difference or / and the dimensional ratio, which is calculated from the dimensional difference / dimension ratio detection unit (FIG. 125 ) and a controller ( 110 ), which is a result of the first comparator ( 108a ) when, in relation to a location determined by the comparison of the first comparator ( 108a ) is determined as a defect, a dimension distribution from the place to a previous place, the dimension difference or / and the dimension ratio by the dimension difference / dimension ratio detection unit (FIG. 125 ), falls within a predetermined range, by comparing the dimension distribution with a dimension distribution in the strip containing the space determined as a defect or a dimension distribution of the chip pattern suspected from the strip in which the dimension difference is in advance affected strip, and a result of the second comparator ( 108b ) instead of the result of the first comparator ( 108a ) stores, when the dimension distribution of the space designated as a defect to the previous place where the dimensional difference or / and the dimensional ratio by the dimension difference / dimension ratio detecting unit (FIG. 125 ) exceeds the predetermined range, by comparing the dimensional distribution with the dimension distribution in the strip containing the defect-determined space or the dimension distribution of the chip pattern suspected from the strip in which the dimensional difference is detected in advance of the stripe concerned.

Further to this aspect of the present invention, an inspection device, wherein the controller ( 110 ) the result of the first comparator ( 108a ) regardless of the dimension distribution from the place designated as a defect to the previous place where the dimension difference or / and the dimension ratio by the dimension difference / dimension ratio detection unit (FIG. 125 ) is determined when the result of the second comparator ( 108b ) does not exist, because the repetitive pattern part does not exist at the place which is determined to be defective, by the comparison of the first comparator ( 108a ).

BRIEF DESCRIPTION OF THE DRAWINGS

1 FIG. 10 is a schematic configuration diagram of an inspection apparatus according to a first embodiment and a second embodiment. FIG.

2 is a view showing a data flow in the inspection device of 1 illustrated.

3 Fig. 13 is a view illustrating a procedure for acquiring an optical image for detecting the defect of the pattern formed in the mask.

4 FIG. 10 is a flowchart illustrating an example of the inspection method according to the first embodiment. FIG.

5 FIG. 12 is a flowchart illustrating an example of the inspection method according to the second embodiment. FIG.

6 illustrates an example of the dimension difference map of the mask.

7 illustrates a dimension distribution corresponding to the map in FIG 6 ,

8th illustrates the dimensional distribution of the pattern measured by the dimension measuring circuit.

9 illustrates the dimensional difference versus the reference value of the line width in the measured part after the influence of the dimension distribution is removed.

10 FIG. 12 is an example of a schematic diagram of a chip pattern of a mask. FIG.

11 is another example of a schematic diagram of a chip pattern of a mask.

12 Fig. 12 is a view explaining a method of measuring a line width, and illustrates a schematic diagram of an optical image of a pattern formed in a mask and a luminance value of each pixel along a broken line.

DETAILED DESCRIPTION OF THE EMBODIMENTS

Embodiment 1

1 FIG. 10 is a schematic configuration diagram of an inspection apparatus according to a first embodiment. FIG. 2 is a view showing a data flow in the inspection device of 1 illustrated. In the 1 and 2 a configuration unit necessary for the first embodiment is illustrated. However, another prior art configuration unit necessary for inspection may be used. As used herein, a "unit" or "circuit" may be configured by a program running in a computer. Alternatively, the "unit" or "circuit" may be constructed not only by the program being software, but also by a combination of software, hardware or firmware. In the case that the "units" or "circuits" are constructed by the program, the program may be recorded on a recording device such as a magnetic disk drive.

In the first embodiment, a mask used in photolithography is considered to be a Inspection target used. Alternatively, as the inspection target in another example, a wafer may be used.

As in 1 illustrated, includes an inspection device 100 includes a configuration unit A constituting an optical image detection unit and a configuration unit B performing processing necessary for inspection using an optical image captured by the configuration unit A.

The configuration unit A includes a light source 103 , an XYθ table 102 which is movable in a horizontal direction (X direction and Y direction) and a rotation direction (θ direction), an illumination optical system 170 forming a transmission illumination system, a magnification optical system 104 , a photodiode array 105 , a sensor circuit 106 , a laser length measuring system 122 and a car charger 130 ,

In the configuration unit A, the optical image 204 a mask 101 , which becomes an inspection target. The optical image data 204 FIG. 12 is an image of a mask in which a figure pattern is written based on design pattern data of the mask 101 contained graphics data. For example, the optical image data 204 8-bit data without code, and express a gradation of the brightness of each pixel.

Several chip patterns are in the mask 101 educated. 10 Fig. 16 is a partially enlarged schematic diagram illustrating the chip pattern. As in 10 Illustrated, 2n are chip patterns in a region in the mask 101 and cells A and B, all formed by a repetitive pattern, are formed in each chip pattern.

The car charger 130 isolated the mask 101 on the XYθ table 102 , An autoloader control circuit 113 drives the car charger 130 under control of a control computer 110 at. If the mask 101 on the XYθ table 102 are positioned in the mask 101 formed pattern with light from the light source 103 that over the XYθ table 102 is arranged, irradiated. The mask becomes more precise 101 irradiated with a light coming from the light source 103 is emitted via the illumination optical system 170 , The magnifying optics system 104 , the photodiode array 105 and the sensor circuit 106 are under the mask 101 arranged. That through the mask 101 transmitted light forms the optical image on the photodiode array 105 via the magnification optical system 104 ,

The magnifying optics system 104 may be configured so that a focus is automatically adjusted by an automatic focusing mechanism (not illustrated). Although not illustrated, the inspection device 100 the mask 101 irradiate light from below and direct the reflected light to the photodiode array via the magnification optical system. In this case, the optical image formed by the transmitted light and reflected light can be simultaneously detected.

The photodiode array 105 performs photoelectric conversion on the pattern image of the mask 101 that on the photodiode array 105 is formed by, and the sensor circuit 106 performs A / D (analog / digital) conversion on the pattern image. A plurality of sensor pixels (not illustrated) are in the photodiode array 105 arranged. A TDI-Ttime Delay Integration (Time Delay Integration) sensor can be given as an example of the sensor. In this case, the TDI sensor captures the image of the pattern in the mask 101 while the XYθ table 102 continuously moving. At this point form the light source 103 , the magnifying optics system 104 , the photodiode array 105 and the sensor circuit 106 a high magnification inspection optical system.

In the configuration unit B is the control computer 110 that is, the whole of the inspection device 100 controlling control, with a position measuring circuit 107 , a comparison circuit 108 , which is a first comparator 108a and a second comparator 108b contains, a reference image generating circuit 112 , which is an example of the reference picture generating unit, a pattern generating circuit 111 an alignment measurement circuit 125 , which is an example of the dimensional difference / dimension ratio detection unit, of a card generating circuit 126 , which is an example of the dimension distribution detection unit, of an autoloader control 113 , a table control circuit 114 , a magnetic disk drive 109 , which is an example of the storage device, a magnetic tape device 115 , a flexible disk drive 116 , a CRT 117 , a sample monitor 118 and a printer 119 over a bus 120 connected, which forms a data transmission line. The XYθ table 102 is controlled by X-axis motor, a Y-axis motor and a θ-axis motor under control of the table control circuit 114 driven. For example, an air slider, a linear motor, and a stepping motor may be used as these driving mechanisms, and further any combination may be used with each other.

As described above, the "unit" or "circuit" in 1 be configured as a program that works on the computer. Alternatively, the "unit" or "circuit" may be constructed not only by the program, which is software, but also by a combination of software, hardware or firmware. In the case where The "unit" or "circuit" can be constructed by the program, the program can drive on the magnetic disk 109 be recorded. For example, each of the autoloader control circuits 113 , the table control circuit 114 , the comparison circuit 108 , and the position measuring circuit 107 be constructed by an electrical circuit, the software being controlled by the control computer 110 can be processed, or the combination of the electrical circuit and the software.

The control computer 110 controls the table control circuit 114 for driving the XYθ table 102 , The movement position of the XYθ table 102 is through the laser length measuring system 122 measured and to the position measuring circuit 107 Posted.

The control computer 110 controls the car charger control circuit 113 to the car charger 130 drive. The car charger 130 automatically promotes the mask 101 , notifies the operator of an end of the inspection, reports a defect as needed, and unloads the mask 101 automatically.

The design pattern data, which will become reference data of the die-to-database process, will be on the magnetic disk drive 109 saved. As the inspection progresses, the design pattern data is read and sent to the pattern generation circuit 111 Posted. The design sample data will be referred to 2 described.

As in 2 Illustrated are CAD data generated by a designer (user) 201 in design intermediate data 202 transformed into a hierarchical format, such as OASIS. The design pattern data generated in each layer and formed in the mask is included in the design intermediate data 202 saved. At this point, the inspection device is generally configured not to read OASIS data directly. That is, independent format data is used by each manufacturer of an inspection device. For this reason, the OASIS data are sent to the inspection device 100 after conversion into format data 203 , which are unique to the inspection device in each layer entered. In this case, the format data 203 be set to a data format suitable for the inspection device 100 is unique or the data format compatible with a drawing device.

The format data 203 be in 1 on the magnetic disk drive 109 entered. That is, the design pattern data obtained during the formation of the pattern in the mask 101 used in the magnetic disk drive 109 get saved.

The figure patterns included in the design pattern may be a rectangle or a triangle as a base graphic pattern. For example, graphics data, in which shape, size and position of each figure pattern on the magnetic disk drive 109 saved. For example, the graphic data is information such as a coordinate (x, y) from the starting position of the graphic pattern, a page length, and a figure code that becomes an identifier identifying a figure pattern type such as a rectangle or a triangle.

A set of graphic patterns existing within a range of several 10 μm is generally called a cluster or a cell, and the data is layered using the cluster or the cell. In the cluster or the cell, a disposition coordinate and a repeating amount are defined in the case where different graphic patterns are arranged separately or repeatedly arranged at a certain distance. The clusters or cell data are arranged in a stripe-shaped area called a stripe. The strip-shaped region has a width of several 100 μm and a length of about 100 mm, which is the total length in the X-direction or Y-direction of the mask 101 equivalent.

The pattern generating circuit 111 reads the input design pattern data from the magnetic disk drive 109 via the control computer 110 ,

In the pattern generation circuit 111 The design pattern data is converted into image data (bit pattern data). That is, the pattern generating circuit 111 extracting the design pattern data into individual data of each graphic pattern and interpreting the figure pattern code and figure pattern dimension indicating the figure pattern shape of the design pattern data. The design pattern data is extracted into binary or multi-level image data as the pattern arranged in a square having a unit of a raster of a predetermined quantization dimension. Then, a busy rate of the graphic pattern in design patterns in each area (square) corresponding to a sensor pixel is calculated, and the occupancy rate of the graphic pattern in each pixel becomes a pixel value.

The through the pattern generating circuit 111 converted image data is sent to the reference image generation circuit 112 that is, the reference image generation unit, and used to produce a reference image (also referred to as reference data).

The from the sensor circuit 106 output program memory optical image data 204 be together with data which is a position of mask 101 on the XYθ table 102 specify to the comparison circuit 108 Posted. The data is taken from the position measuring circuit 107 output. The reference image is also sent to the comparison circuit 108 Posted.

In the comparison circuit 108 become the optical image data 204 and comparing the reference data using a suitable comparison determination algorithm. In the configuration of 1 Transfer images are compared. In a configuration in which a reflection optical system is used, reflection images are compared with each other, or a comparison determination algorithm in which transmission and reception are combined is used. As a result of the comparison, in a case where a difference between the two exceeds a predetermined threshold, the position is determined as a defect.

For example, the determination threshold registered as a line width defect becomes a line width dimension difference (nm) between the optical image data 204 and the reference data and a dimension ratio (%). For example, the determination thresholds are assigned to the line width dimension difference of 16 nm and the aspect ratio of 8% in two ways. When the dimension difference with the reference data is 20 nm while the pattern of the optical image data 204 has a line width of 200 nm because the pattern is larger than both the thresholds of the dimensional difference and the dimensional ratio, the pattern is registered as the defect.

In the case where the line width is larger than that of the reference data and the case where the line width is smaller than that of the reference data, the threshold value of the defect determination may be assigned separately. The threshold value can be assigned to the case where the inter-pattern distance is larger than that of the reference data in both the case where the inter-pattern distance is not greater than the case where the inter-pattern distance is smaller than that of the reference data. The threshold values of a hole diameter or a diameter-dimension ratio may be assigned to the pattern having a hole shape. In this case, the threshold may be assigned to sections in the X direction and Y direction of the hole.

In the comparison circuit 108 becomes the (stripe-shaped) optical image data 204 corresponding reference image divided into small rectangular regions of several 10 microns, which are called inspection frames. A sensor frame image resulting from the optical image data 204 is extracted, and a reference frame image extracted from the reference image are input to a comparison unit. The comparison unit compares the sensor frame image and the reference frame image with each other to detect the defect. Several ten comparison units are in the comparison circuit 108 included, so as to simultaneously process several inspection frames. Each comparison unit detects the unprocessed frame image when finishing the processing of an inspection frame. Therefore, many inspection frames are processed sequentially.

The processing of the comparison unit is specifically carried out as follows. The sensor frame image and the reference frame image are aligned with each other. At this time, to align edge positions of the pattern or luminance peak positions, the sensor frame image or the reference frame image is shifted in parallel in units of sensor pixels and becomes the sensor frame image and the reference frame image up to the sensor pixel or less by flipping luminance values aligned to adjacent pixels. After alignment, a level difference between the sensor frame image and the reference frame image in each pixel is evaluated, and derivative values of the pixels in a pattern edge direction are compared with each other, whereby the defect is detected according to the proper comparison algorithm. Hereinafter, the comparison of the sensor frame image and the reference frame image is sometimes referred to simply as a comparison of the optical image and the reference image. The sensor frame images are compared with each other in the comparison by the die-to-die method. However, in this case, sometimes the comparison of the sensor frame images is referred to simply as the comparison of the optical images.

At the same time, in the comparison circuit 108 according to the repetitive pattern in the optical image data 204 and they are extracted with a proper range of dimensions, and the cells are compared. The repetitive pattern is determined by a method of extracting a pattern feature from a repetitive disposition command of the graphic pattern or the cell included in the layered structure in the design data and the acquired optical image data 204 are included, searched. For example, a subframe that becomes a unit of a predetermined repetitive pattern is defined in the inspection frame of the optical image. The subframes are compared with each other in only one inspection frame, and the repetitive pattern is determined to have the defect when a difference exists between the patterns of the subframe.

In the first embodiment, the optical image data also becomes 204 to the alignment measurement circuit 125 Posted. In the alignment measurement circuit 125 for example, the line width of the mask 101 written line pattern from the optical image data 204 measured. The reference image generation circuit 112 sends the reference data to the alignment measurement circuit 125 and the position measuring circuit 107 sends the, the position of the mask 101 on the XYθ table 102 indicative data to the alignment measurement circuit 125 , In the alignment measurement circuit 125 the line width is measured according to the line pattern from that from the reference data. The dimensional difference or the dimension ratio between the pattern line width of the optical image and the pattern line width of the reference image is determined based on the measured value.

The pattern dimension measurement in the alignment measurement circuit 125 becomes simultaneous with the capture of the optical image by the mask 101 carried out. Alternatively, for example, the pattern dimension measurement in the alignment measurement circuit 125 simultaneously with that by the comparison circuit 108 performed inspection.

The alignment measuring circuit 125 FIG. 10 is an example of the dimensional difference / dimension ratio detection unit of the invention. FIG. In the embodiment, the space width between the line patterns, namely, the interline distance instead of the line width, is measured, and the difference or the ratio between the inter-line distances can be obtained. Both the dimension difference and the dimension ratio between the line widths or the interline distances can be determined.

In the dimension difference or the dimension ratio, for example, the pattern of the mask 101 divided to form a plurality of inspection areas, and the line width of each pixel for the optical image of each inspection region is determined. Then, a frequency of the determined line width is compiled, and an average value of the line widths is calculated from the compiled result of the frequency distribution. The dimensional difference or the dimension ratio of the line width is determined from the average value and the line width determined from the reference image. Specifically, that can be found in the Japanese Patent No. 3824524 disclosed methods are applied.

The data of the dimensional difference or the dimension ratio caused by the alignment measuring circuit 125 is detected is sent to the card generating circuit 126 that is, the dimension distribution detection unit. In the card generating circuit 126 For example, a map of the dimensional difference or the dimension ratio of the pattern line width on the surface of the mask will be 101 produced based on the data sent. The generated card will be on the magnetic disk drive 109 saved. The inspection device 100 does not necessarily have the card generating circuit 126 included, but the alignment measurement circuit 125 may have the card generating function or the card may be generated by an external computer. Alternatively, a defect determination may be made by the data of the dimensional difference or the dimension ratio, which is determined by the alignment measurement circuit 125 is determined without generating the card.

When the comparison circuit 108 if the pattern is found to have a defect, the coordinate of the defect and the optical image and reference image which are the basis of the defect determination become as a mask inspection result 205 on the magnetic disk drive 109 saved. The mask inspection result 205 becomes a review tool 500 sent as in 2 illustrated. A review process is an operation in which the operator determines whether the detected defect becomes a practical problem. For example, the operator visually determines whether the defect needs to be corrected by comparing the reference image that is the basis of the defect determination with the defect-containing optical image.

The defect information determined through the review process also becomes on the magnetic disk drive 109 from 1 saved. As in 2 illustrated when the defect to be corrected by the review tool 500 is confirmed, the mask becomes 101 to a repair device 600 sent, that is the external device of the inspection device 100 together with a defect information list 207 , Since a correction method depends on whether the defect is protruding or recessed, a defect type that is the difference between protrusion and depression and the defect coordinate becomes the defect information list 207 added.

An example of a method for inspecting the mask 101 with the inspection device 100 in 1 is described below.

4 FIG. 10 is a flowchart illustrating an example of the inspection method of the first embodiment. FIG. As in 4 As illustrated, an inspection process includes a process of determining the optical image of the mask 10 (Optical Image Acquisition Process; S1), a process of storing the design pattern data of the in the mask 101 formed pattern (memory design Pattern data; S2), a pattern generation process (S3), that is, an example of the process of generating the reference image (S3) and a filtering process (S4), a process of comparing the optical image with the image that becomes a reference (comparison process; S6), a process of measuring the pattern dimension difference from the optical image and the reference image (dimension measuring process; S7), a process of generating the dimension difference map in the surface of the mask 101 based on the measured dimensional difference (map generating process; S8), a process of comparing the dimension distribution in the vicinity of the defect detection position to the dimension distribution in another region (S9), and a process of determining whether the dimension distribution in the neighborhood of the defect Detection position falls within a predetermined range (S10).

(Optical image acquisition process)

The mask 101 gets on the XYθ table 102 positioned. In order to determine the correct inspection result, it is necessary to use the mask 101 at a predetermined position of the XYθ table 102 to locate. Therefore, usually an alignment mark in the mask 101 formed and the mask 101 gets on the XYθ table 102 aligned using the alignment mark.

The orientation of the mask 101 (Plate alignment) is with reference to 11 described. In the example of 11 are cross mask alignment marks MA at four corners of the mask 101 educated. Several chip patterns (C1, C2, C3, ...) are in the mask 101 and chip alignment marks CA are also formed in each chip. The mask 101 gets on the XYθ table 102 positioned and it is assumed that the XYθ table 102 a horizontally moving XY stage and a θ stage disposed on the XY stage to move in a rotational direction.

Specifically, in the alignment process, an X-axis and a Y-axis of the pattern, that is, the inspection target, are aligned with running axes of the XY stage while the mask 101 on the XYθ table 102 is positioned.

In the mask alignment marks MA provided at the four positions, the images of the two mask alignment marks MA having the smaller values of the Y coordinates are detected, the θ stage is rotated so that the two mask alignment marks MA the be the same Y-coordinate correctly, reducing the direction of rotation of the mask 101 fine adjusted. At this point, the distance between the mask alignment marks MA is correctly measured. Then, the images are captured with the two mask alignment marks MA having the larger values of the Y coordinates. Therefore, the coordinates of the mask alignment marks MA are correctly measured at four positions.

It is found from the above measurement that the two mask alignment marks MA having the larger values of the Y coordinates are located at vertices of a trapezoid, the two mask alignment marks MA having the smaller values of the Y coordinates at both ends become a base exhibit. At this point, because the mask 101 has originally a rectangular shape, it is assumed that the two mask alignment marks MA having the larger values of the Y coordinates are located at the corner points of the rectangle. However, the measurement result shows that the two mask alignment marks MA with the larger values of the Y coordinates are located at the corner points of the trapezoid. Taking into account these facts, it is discovered that the shape of the mask 101 is deformed. Accordingly, it is assumed that the region of the pattern becomes the inspection target having trapezoidal deformation similar to the trapezoid and expansion and contraction of the distance between the mask alignment marks MA, and the compensation becomes on the assumption of trapezoidal deformation and the expansion and contraction of the distance performed when the reference image generating circuit 112 generates the reference data.

The mask alignment mark MA is not necessarily in the mask 101 intended. In this case, alignment is performed using the vertex of the corner or the side of the edge pattern that is close to the outer periphery of the mask 101 is equal to the coordinates of the XY coordinate in the pattern which becomes the inspection target.

In the optical image detection process of the first embodiment, the configuration unit A in FIG 1 the optical image of the mask 101 , 3 FIG. 13 is a view illustrating an optical image detection procedure for the purpose of detecting the defect of the mask 101 Illustrated pattern illustrated. As described above, the optical image corresponds to the optical image data 204 in 2 ,

In 3 it is assumed that the mask 10 on the XYθ table 102 in 1 is positioned. The inspection region in the mask 101 becomes virtually strip-shaped multiple inspection regions, namely stripes 20 ₁ , 20 ₂ , 20 ₃ , 20 ₄ , ... divided, as in 3 illustrated. For example, each strip is in a region having a width of several hundred micrometers and a length of about 100 mm corresponding to the total length in the X direction or Y direction of the mask 101 ,

The optical image is captured in each strip. That is, when capturing the optical image in 3 the operation of the XYθ table 102 is controlled so that each strip 20 ₁ , 20 ₂ , 20 ₃ , 20 ₄ , ... is scanned continuously. Specifically, the optical image of the mask 101 recorded while the XYθ table 102 in the -X direction of 3 moving. The image that has a scan width W in 3 is continuous at the photodiode array 105 in 1 entered. That is, the image of the second strip 20 _{2 is} captured after the image of the first strip 20 _{1 is} detected. In this case, after the XYθ table 102 moving in the -Y direction in a stepwise manner so that the optics image is detected while the XYθ stage is being detected 102 in the direction (X-direction) opposite to the direction (-X-direction) in which the image of the first strip moves 20 ₁ , and the image having the scan width W continuously at the photodiode array 105 is entered. In the event that the picture of the third strip 20 ₃ , after moving in the -Y direction in a stepwise manner, the XYθ table is detected 102 in the direction opposite to the direction (X-direction) in which the image of the second stripe is moving 20 ₂ , that is, the direction (-X direction) in which the image of the first stripe is detected 20 _{1 is} detected. An arrow in 3 indicates the optical image detection direction and sequence, and a hatched area indicates the area where the optical image is already detected.

10 illustrates a state in which the optical image is being captured. In 10 2n chip patterns in a predetermined region in the mask 101 and a cell A and a cell B, each containing the repetitive pattern, are formed in each chip pattern. The sensor detects the image of the pattern along the strip in the order of the first chip, the second chip, the third chip,..., And the nth chip.

The photodiode array 105 performs photoelectric conversion on the photodiode array 105 in 1 formed pattern image and the sensor circuit 106 performs an A / D (analog-to-digital) conversion on the pattern image.

Then the optical image from the sensor circuit 106 to the comparison circuit 108 in 1 Posted.

The A / D converted sensor data is input to a digital amplifier (not illustrated) that can adjust an offset and a gain in each pixel. The gain for each pixel of the digital amplifier is set in a calibration process. For example, in the transmitted light calibration process, a black level is set while the image is a light-shielding region in the mask 101 that is sufficiently wide with respect to an area in which the image is detected by the sensor. Then, a white level is set while the image of a transmitted bright region in the mask 101 which is sufficiently wide with respect to an area in which the image is picked up by the sensor. At this point, considering fluctuation in the amount of light during the inspection, the offset and the gain in each pixel are adjusted so that the amplitudes of the white level and the black level in a range of 10 to 240 corresponding to about 4% to 94% of 8-bit gradation data are distributed.

(Storage process)

In the case of inspection by the die-to-database comparison method, the reference image generated from the design pattern data becomes a reference of a defect determination. In the inspection device 100 become the for forming the pattern in the mask 101 used design pattern data on the magnetic disk drive 109 saved.

(Pattern forming process)

In the pattern generation process, the pattern generation circuit reads 111 in 1 the design pattern data from the magnetic disk drive 109 via the control computer 110 and converts the read design pattern data of the mask 101 into the binary or value-added image data (design image data). The image data is sent to the reference image generation circuit 112 Posted.

(Filtering process)

In the filtering process, the reference image generating circuit performs 112 in 1 the proper filtering on the design pattern data, that is the graphic image data. The reason is as follows.

In the production process, because the roundness of the corner and a finished dimension of the line width is adjusted, the pattern is in the mask 101 not strictly according to the design pattern. The optics image data 204 that is, from the sensor circuit 106 in 1 obtained optical image is due to a resolution characteristic of the magnification optical system 104 or an aperture effect of the photodiode array 105 weak, in other words in a state in which a spatial low-pass filter works.

Therefore, the mask that becomes the inspection target is observed before inspection becomes Filter coefficient, which mimics the manufacturing process or a change of an optical system of the inspection device, determined to subject the design pattern data to a two-dimensional digital filter. Thus, the processing of imitating the optical image on the reference image is performed.

The learning process of the filter coefficient may be fixed by using the pattern of the mask which becomes the reference in the production process or in the part of the pattern of the mask (in the first embodiment, mask 101 ), which becomes the inspection target. In the latter case, the filter coefficients are determined in consideration of the pattern line width of the region used in the learning process or a finished degree of roundness of the corner and reflected in a defect determination criterion of the overall mask.

In the case where the mask which becomes the inspection target is used, the learning process of the filter coefficient can be advantageously performed without removing influences such as variation of the production lot or fluctuation in the condition of the inspection apparatus. However, if the dimension on the surface of the mask fluctuates, the filter coefficient with respect to the position used in the learning process becomes optimal, but the filter coefficient does not necessarily become optimal with respect to other positions, resulting in a pseudo-defect , Therefore, preferably, the learning process is performed around the surface center of the mask, which is hardly affected by the fluctuation in the dimension. Alternatively, the learning process is performed at a plurality of positions on the surface of the mask, and the average value of the determined plural filter coefficients can be used (dimension measuring process).

In the dimension measuring process, the pattern dimension difference is measured from the optical image and the reference image. In the in 1 shown inspection device 100 Measures the alignment measurement circuit 125 the dimensional difference of the pattern line width between the optical image and the reference image using the sensor circuit 106 output optical reference data 204 and that from the reference image generation circuit 112 output reference data. The dimension ratio of the pattern line width may be measured instead of or in addition to the dimension difference of the pattern line width, or the inter-pattern distance difference or the inter-pattern distance ratio may be determined instead of or in addition to the pattern line width.

For example, a frequency at which the alignment measurement circuit 125 measures the dimensional difference during the inspection, to a proper number of scanning times (about 1000 points) in the length direction (X direction) of the strip ( 20 ₁ , 20 ₂ , 20 ₃ , 20 ₄ , ...) in 3 and set to almost the same number of scanning times in the width direction (Y direction) of the strip. A proper line pattern in which a distance of an edge pair can be measured is used in the vicinity of a potential point at which the dimension difference is measured. In this case, one pair of edges can be used. However, preferably, the dimension difference is measured using the edge pairs of plural positions, the frequency of the detected value is compiled, and the highest frequency value (mode) of the compile result of the frequency distribution is used as a representative value. In the case where the edge pair is not found in the vicinity of the potential point, or in the case of a small number of edge pairs, the dimension difference need not be measured or the mode can be determined from the limited number of samples.

(Map generation process)

In the inspection device 100 measures at the same time, if the optic image of the mask 101 is detected, the alignment measuring circuit 125 the pattern dimension difference between the optical image and the reference image and become the obtained data of the dimensional difference to the card generating circuit 126 Posted. In the card generating circuit 126 For example, the map expressing the dimension distribution on the surface of the mask is generated from the accumulated data of the dimensional difference. The dimension distribution in the currently inspected strip or the dimension distribution of the strip in which the inspection has already been performed in the same mask can be recognized from the map.

(The database comparison process and cell comparison process)

As in 2 illustrated, the optical image data acquired in the optical image detection process becomes 204 to the comparison circuit 108 Posted. The reference image generation circuit 112 sends the reference data to the comparison circuit 108 , The comparison circuit 108 contains the first comparator 108a and the second comparator 108b and the first comparator 108a compares the optical image data 204 with the reference data by the die-to-die method. Simultaneously with the processing performed by the die-to-database comparison method, the second comparator searches 108b after the repetitive pattern in the optical image data 204 and extracts the repetitive pattern in an appropriate dimensional range to the Perform cell comparison. However, in the case that the cell which becomes the reference does not exist because the repetitive pattern does not exist near the cell which becomes the inspection target, the processing is performed only by the to-database comparison method.

In both methods, the data that becomes the inspection target and the data that becomes the reference of the defect determination are compared with each other using the correct comparison determination algorithm. The data which becomes the inspection target is detected as a defect in the case where the difference between the two exceeds the predetermined threshold.

For example, it is assumed that the chip pattern in the mask 101 Matrices are aligned. In the die-to-database comparison method, when the n-th chip is adopted as the inspection target, the n-th chip is detected as a defect in the case where the pattern difference between the optical image and the reference image of the n-th Chips exceeds the predetermined threshold. On the other hand, in the cell comparison method, the patterns which are separated from each other by a distance of the repetitive pattern (cell), such as the memory matte function in a chip, are compared with each other, and the pattern is detected as a defect in the case where the difference between the both exceed the predetermined threshold. In this case, when the specific cell in the nth chip is the inspection target, the optical image preceding the specific cell becomes the reference image to be compared. Assuming, for example, that the second cell is the inspection target in cell A of FIG 10 is, the optical image of the first cell is the reference image.

More specifically, the defect determination can be made by the following two methods. One of the methods is the method for determining that the inspection target is the defect in the case where the difference exceeding a predetermined threshold between the position of a contour in the reference image and the position of a contour in the optical image is recognized. The other method is the method for determining that the inspection target is the defect in the case where the ratio of the pattern line width in the reference image and the pattern line width in the optical image exceeds a predetermined threshold. In this method, the ratio of the inter-pattern distance in the reference image and the inter-pattern distance in the optical image can be used.

(Process of determining whether the dimension distribution in the vicinity of the defect detection position falls within a predetermined range by comparing the dimension distribution in the vicinity of a defect detection position with the dimension distribution of another region)

In the case where the alignment measuring circuit 125 the pattern dimension simultaneously with the detection of the optical image of the mask 101 is measured on the most recent data in the by the alignment measurement circuit 125 measured dimension difference data when the defect is detected by the die-to-database comparison method. When the dimensional distribution from the position where the defect is detected to the previous position where the dimensional difference is detected falls within the predetermined range, by comparing the dimensional distribution with the dimension distribution in the chip and the dimension distribution between the chips, the result of FIG -to-database comparison method, that is, the defect coordinate and the optical image and reference image, which form the basis of the defect determination, as the mask inspection result 205 in the magnetic disk drive 109 saved.

In the case where the alignment measuring circuit 125 the pattern dimension simultaneously with the inspection of the comparison circuit 108 at a time when the defect is detected by the die-to-database comparison method, the dimension of the pattern where the comparison has already been made is measured, but the dimension of the pattern in which the comparison is not performed is measured is not measured. Therefore, in this case, the most recent data is taken from the dimensional difference data obtained by the alignment measuring circuit 125 be measured.

On the other hand, the defect is detected by the die-to-database comparison method, and the dimension distribution from the position where the defect is detected to the previous position where the dimensional difference is detected exceeds the predetermined range by comparing the dimension distribution with the dimension distribution in the chip and the dimension distribution between the chips. In this case, the result of the die-to-database comparison method is not used in relation to the position, but the result of the concurrent cell comparison method is used. At this point, whether the defect is detected, as a result of the cell comparison process, no problem. That is, even if the position determined as a defect by the die-to-database comparison method is not as the defect is registered unless the position is determined to be defective by the cell comparison method.

However, in the case where the cell which becomes the reference does not exist because the repetitive pattern does not exist near the cell which becomes the inspection target, the processing is performed only by the die-to-database comparison method. In this case, even if the dimensional distribution from the position where the defect is detected to the previous position where the dimensional difference is detected exceeds the predetermined range, by comparing the dimensional distribution with the dimensional distribution in the chip and the dimension distribution between the chips preferably the result of the die-to-database comparison process. That is, the coordinate of the defect detected by the die-to-database comparison method and the optical image and reference image that are the basis of the defect determination become a mask inspection result 205 on the magnetic disk drive 109 saved.

By way of example, the dimensional distribution from the position where the defect is detected to the previous position where the dimensional difference is detected is compared with the dimensional distribution in the chip and the dimension distribution between the chips. Alternatively, the dimension distribution from the position where the defect is detected to the previous position where the dimensional difference is detected may be compared with the dimension distribution of the region separated by the chip gap.

For example, when the dimensional distribution from the position where the defect is detected to the previous position where the dimensional difference is detected falls within the predetermined range, by comparing the dimension distribution with (1) the dimensional distribution in the position determined as the defect position, or (2) the dimension distribution of the chip pattern adopted from the strip in which the dimension difference is detected in advance from the strip containing the position determined as a defect, the defect coordinate and the optical image and reference image which are the The basis of the defect determination are, on the magnetic disk drive 109 get saved. At this point, (1) the dimension distribution and (2) the dimension distribution are calculated by the map generating circuit 126 derived map derived. (1) the dimensional distribution and (2) the dimensional distribution can also be derived directly from the dimensional difference data obtained by the alignment measuring circuit 125 be determined.

In the above modification, the defect is detected by the die-to-database comparison method, and the dimension distribution from the position where the defect is detected to the previous position where the dimensional difference is detected is given by (1) the dimension distribution and ( 2) of the dimension distribution. When the dimension distribution deviates from the predetermined range from the position where the defect is detected to the previous position where the dimensional difference is detected, the result of the die-to-database comparison method is not adopted with respect to the position, but becomes used the result of the parallel cell comparison procedure. In this case, whether the defect is detected as a result of the cell comparison method is not a problem. That is, even the position determined as a defect by the die-to-database comparison method is not registered as the defect unless the position is determined to be defective by the cell comparison method.

However, in the case where the cell which becomes the reference does not exist because the repetitive pattern does not exist near the cell which becomes the inspection target, the processing is performed only by the die-to-database comparison method. In this case, even if the dimensional distribution deviates from the position where the defect is detected to the previous position where the dimensional difference is detected is deviated from the predetermined range, by comparing the dimensional distribution with (1) the dimensional distribution and (2) the dimension distribution, the result of the die-to-database comparison process used. That is, the coordinate of the defect detected by the die-to-database comparison method and the optical image and reference image that are the basis of the defect determination are referred to as the mask inspection result 205 in the magnetic disk drive 109 get saved.

In the case where the defect is detected by the die-to-database comparison method, the excess of the determination distribution is changed from the position where the defect is detected to the previous position where the dimensional difference exceeds the predetermined range by comparing the difference As the dimensional distribution in the chip and the dimension distribution between the chips is determined, the tendency of the line width in the region where the learning process of the two-dimensional digital filter is performed differs from the line width tendency at the position where the defect is generated during the generation Reference data is detected. The same applies to the case where the dimension distribution from the position where the defect is detected to the previous position where the dimensional difference is detected is the predetermined one Range exceeds, by comparing the dimensional distribution with (1) the dimension distribution in, the strip containing the defect detected position, or (2) the dimension distribution of the chip pattern adopted from the strip in which the dimension difference is determined beforehand, inclusive the position determined as a defect.

Therefore, in the first embodiment, whether the reference data for the reference of the defect determination is suitable by comparing the tendency of the dimensional difference in the chip, the tendency of the dimension between the chips, the tendency of the dimensional difference in the same chip or the tendency of the dimensional difference in the chip Surface of the mask with the tendency of the dimension distribution from the position where the defect is detected, to the previous position where the dimensional difference is determined, are determined. The determination can be made by the control computer 110 in 1 be made. The control computer 110 determines whether the result of the cell comparison process exists, and the control computer 110 Sets the result of the die-to-database comparison process to the mask inspection result 205 irrespective of the result of the comparison in the case that there is no result of the cell comparison method, but only the result of the die-to-database comparison method.

In the case where the dimension distribution from the position where the defect is detected to the previous position where the dimensional difference is detected exceeds the predetermined range as a result of the comparison because the dimension in the surface of the mask fluctuates, the filter coefficient is the defect detection position is not the optimum value, and the reference data is determined not to be suitable for the reference of the defect determination. The result of the cell comparison procedure is assumed. If the position is also determined to be defective by the cell comparison method, the position is registered as the defect. For example, the control computer saves 110 the defect coordinate and the optical image and reference image, which are the basis of the defect determination, as the mask inspection result 205 in the magnetic disk drive 109 , Unless the position is determined as a defect by the cell comparison method, it is considered that the defect detected by the die-to-database comparison method falls within an acceptable range, but the position is not registered as the defect. However, the result of the die-to-database comparison process can be stored.

When passing through the alignment measuring circuit 125 determined dimension difference is large, the position can be registered as the defect, even if the position is not considered the defect in the comparison circuit 108 is determined. Therefore, in the first embodiment, the defect determination can be made as follows.

For example, it is assumed that the threshold value by which the comparison circuit 108 determines the line width defect to which the line width dimension difference of 16 nm and the aspect ratio of 8% are set. The threshold value of the defect determination for the measurement result of the alignment measurement circuit 125 is compared to the threshold value by which the comparison circuit 108 Determines the line width defect, relaxes slightly and the dimension ratio is set to 10%. With respect to the predetermined range which becomes the criterion for which the result of the die-to-database comparison method or the result of the cell comparison method is adopted, the line width dimension difference is set to 12 nm or larger and a dimension ratio is set to 6%. or larger.

The result of the die-to-database comparison method is used when passing through the alignment measurement circuit 125 determined dimensional difference is smaller than 12 nm, while the dimensional ratio is smaller than 6%. On the other hand, the result of the cell comparison method is used when passing through the alignment measurement circuit 125 The measured dimensional difference is greater than or equal to 12 nm and less than 20 nm, while the dimensional ratio is greater than or equal to 6% and less than 10%. The position is registered as the defect when passing through the alignment measurement circuit 125 The measured dimensional difference is greater than or equal to 20 nm while the dimensional ratio is greater than or equal to 10%.

The predetermined range, which becomes the criterion for which the results of the die-to-database comparison process and the result of the cell comparison process are adopted, is set in each mask that becomes the inspection target. At this point, the predetermined range is set to the range that does not exceed the threshold in the case where the position is determined as the defect, from the measured value of the alignment measuring circuit 125 , The adjustment method is similar to the threshold adjustment method in the comparison circuit 108 , That is, the predetermined range can be assigned individually for the case where the line width is larger than the reference data and the case where the line width is smaller than the reference data, and the predetermined range can be assigned for the case. where not the line width, but the intermediate pattern distance greater than the Are reference data, and the case where the inter-pattern distance is smaller than the reference data. In addition, the predetermined area of the hole diameter or the dimensional ratio of the diameter may be assigned to the pattern having the hole shape. In this case, the predetermined range may be assigned to both the X-direction and the Y-direction portions of the hole.

In the conventional inspection method, all the information about the defect detected by the die-to-database comparison method is registered. Therefore, sometimes the defect information that was not originally detected is registered. On the other hand, in the first embodiment, the result of the die-to-database comparison method is replaced by the result of the cell comparison method in some cases to remove the influence of the line width distribution on the surface of the mask from the obtained data. Therefore, because the defect that does not have to be originally detected is removed from the mask inspection result, the number of defects viewed by the operator decreases, thereby shortening the inspection time. Also, because the number of defects described in the defect information list decreases, the production yield of the mask can be improved. In addition, the shape defect and the defect caused by the fluctuation in the local line width can be detected by removing the influence of the line width distribution in the surface of the mask.

The through the card generating circuit 126 The card produced in the first embodiment can be used to match the pattern of the mask 101 to transfer to the wafer. For example, if the exposure device containing the pattern of the mask 10 transfers to the wafer, can input irradiation energy (dose) as a card, which is generated by the card generating circuit 126 generated map on the exposure device and converted into a map of the irradiation energy, which allows the line width is transmitted homogeneously to the wafer. For example, at the position where the dimensional difference in the mask 101 is negative, that is, the position where the line width is thinned, the irradiation energy is adjusted so that the pattern transmitted to the wafer is thickened. On the other hand, at the position where the dimensional difference in the mask 101 positively, namely the position where the line width is thickened, the irradiation energy is adjusted so that the pattern transferred to the wafer is thinned. Therefore, the line width of the pattern transferred to the wafer is uniformly homogenized in the mask in which the pattern has the dimension distribution.

Second embodiment

The inspection method in which the die-to-database comparison method and the cell comparison method are combined is described in the first embodiment. In a second embodiment, the inspection may be performed by the combination of the to-the-match method and the cell comparison method. In this case, the inspection device 100 in 1 be used.

The die-to-die comparison method is the method of comparing the optical images of the same pattern in the chips on different masks with each other in the case where the plural chips have a pattern configuration partially or in total, are arranged in the same mask. That is, the chips repetitively formed in the mask are compared with each other in the die-to-die comparison method, and the corresponding patterns, such as the memory mat regions in the one chip, namely the cell, are compared with each other in the cell comparison process. In the example of 10 For example, the first chip and the second chip are compared with each other in the die-to-die comparison method. On the other hand, in the cell comparison method, the first cell and the second cell of the cell A are compared with each other.

The die-to-die comparison method and the cell comparison method will be described in detail with reference to 11 described. As in 11 illustrated, the sensor detects the image of the pattern along the strip 20 in the order of the chip C1, the chip C2 and the chip C3. In the comparison circuit 108 the captured image of the stripe data is divided into units of inspection frames and becomes an inspection frame extracted from the chip C1 1 with an inspection frame extracted from the chip C2 2 compared. The comparison is performed by the comparison unit, which in the comparison circuit 108 intended to carry out the processing for each unit of the inspection framework.

Several ten comparison units are provided so as to be able to simultaneously process a plurality of inspection frames, and each comparing unit detects the unprocessed frame image when the processing of an inspection frame is ended. Specifically, the comparison unit directs the image of the inspection frame 1 to the picture of the inspection frame 2 out. At this point, to align the edge positions of the pattern or the luminance peak positions, the image of the inspection frame becomes 1 or the picture of the inspection frame 2 shifted parallel in units of sensor pixels. The picture of the inspection frame 1 and the picture of the inspection frame 2 become aligned to the sensor pixel or smaller by pre-grading the luminance values of neighboring pixels. Then the first comparator leads 108a the die-to-die comparison on the picture of the inspection frame 1 and the picture of the inspection frame 2 by. Similarly, the die-to-die comparison becomes the image of the inspection frame 2 and the picture of the inspection frame 3 carried out.

As in 11 As illustrated, the subframe that becomes a unit of the repetitive pattern is defined in each inspection frame. For example, the second comparator performs 108b the cell comparison of the subframe 1 and the image of a subframe 2 in the inspection framework 1 out.

In both the die-to-the-compare method and the cell comparison method, the level difference between the two images to be compared in each pixel is evaluated, and the derivative values of the pixels in the pattern edge direction are compared with each other, whereby the defect is determined by the proper comparison algorithm is detected.

5 FIG. 10 is a flowchart illustrating an example of the inspection method of the second embodiment. FIG. It means that 5 illustrates the inspection method in which the to-the-match method and the cell comparison method are combined.

As in 5 illustrated, the inspection process includes a process of detecting the optical image of the mask 101 (Optical Image Acquisition Process S11), a process of comparing the optical image which becomes the inspection target with the optical image (also referred to as a reference image) which becomes the reference (comparison process S12 and S13), a process of measuring the pattern dimensioning Difference from the optical image and the reference image (dimension measuring process S14), a process of generating the dimensional difference map on the surface of the mask 101 Based on the measured dimensional difference (map generating process S15), a process of comparing the dimension distribution in the vicinity of the defect detecting position with the dimension distribution in another region (S16) and a process of determining whether the embodiment in the vicinity of the defect Detection position falls within a predetermined range (S17).

(Optical image detection process)

Because the optical image detection process (S11) in FIG 5 already in the 1 to 4 of the first embodiment, a description will be omitted.

(Dimension measuring process)

In the inspection device 100 Measures the alignment measurement circuit 125 the pattern dimension difference between the optical images parallel to the detection of the optical image of the mask 101 , In the dimension measuring process (S14), the pattern dimension difference between the optical image that becomes the inspection target and the optical image that becomes the reference is measured by the die-to-die comparison method. In the inspection device 100 in 1 Measures the alignment measurement circuit 125 the dimensional difference of the pattern line width between the optical images using the sensor circuit 106 output optical image data.

A specific example of the linewidth measuring method will be described with reference to FIG 12 described. 12 is a view which schematically the optical image of the mask 101 and a luminance value of each pixel illustrated along a broken line. For example, if the line and the space pattern, as in 12 is illustrated, the threshold in the center of the white and black amplitude of the optical image is set to the threshold of the line width determination, and the position of the pattern edge is determined from the pixel position where the luminance value of the optical image crosses the threshold. Then, the positions of the pattern edges are converted to the distance of the edge pair to determine the line width.

In the dimension measuring process, the position information of the mask becomes 101 on the XYθ table 102 to the optical image data from the sensor circuit 106 added. The position information becomes out of the position measuring circuit 107 Posted. The dimension ratio of the pattern line width may be measured instead of or in addition to the dimension difference of the pattern line width, or the inter-pattern distance difference or the inter-pattern distance ratio may be determined instead of or in addition to the pattern line width.

For example, the frequency at which the alignment measurement circuit 125 measures the dimensional difference during the inspection to the correct number of sampling times (about 1000 points) in the length direction (X direction) of the strip ( 20 ₁ , 20 ₂ , 20 ₃ , 20 ₄ , ...) in 3 are set and the approximately equal number of scanning times in the width direction (Y direction) of the strip are set. The correct line pattern, at which the distance of the edge pair can be measured, is used in the neighborhood of the potential point where the dimension difference is measured. In this case, this can be an edge pair be used. However, preferably, the dimensional difference is measured using the edge pairs of plural positions, the frequency of the detected value is added, and the highest frequency value (mode) of the compiled result of the frequency distribution is used as a representative value. In the case that the edge pair is not found in the vicinity of the potential point, or in the case of a small number of edge pairs, the dimensional difference need not be measured or the mode can be determined from the limited number of samples.

(Map generation process)

The data sent by the alignment measurement circuit 125 determined dimension difference are sent to the card generating circuit 126 Posted. In the card generating circuit 126 the map expressing the dimension distribution on the surface of the mask is generated from the accumulated data of the dimensional difference (map generating process S15). The dimension distribution in the currently inspected strip or the dimensional difference of the strip in which the inspection is already performed in the same mask can be recognized from the map. (The-to-the-comparison process and cell comparison process).

The optical image data captured in the optical image acquisition process 204 be to the comparison circuit 108 in 1 Posted. The comparison circuit 108 includes the first comparator 108a and the second comparator 108b , In the second embodiment, the first comparator compares 108a the optical image data by the die-to-die method with each other (the to-the-comparison process S12). The second comparator searches in parallel with the comparison made by the compare-to-the-comparison process 108b the repetitive pattern in the optical image data 204 and extracts the repetitive pattern in the proper size range to perform the cell comparison (cell comparison process S13). However, in the case where the cell which becomes the reference does not exist because the repetitive pattern does not exist near the cell which becomes the inspection target, the processing is performed only by the to-the-die comparison method.

In both methods, the data that becomes the inspection target and the data that becomes the reference of the defect determination are compared with each other using the correct comparison determination algorithm. The data that becomes the inspection target is determined as a defect in the case where the difference between the two exceeds the predetermined threshold.

For example, assume that the grid-shaped chip pattern in the mask 101 Matrix-aligned. In the die-to-die comparison method, when the n-th chip is considered as the inspection target, the n-th chip is determined to be the defect in the case where the pattern difference between the optical image of the n-th chip and the n-th chip Optics image of the (n-1) -th chip exceeds the predetermined threshold. On the other hand, in the cell comparison method, the patterns separated from each other by the pitch of the repetitive pattern (cell) such as the memory mat area in a chip are compared with each other, and the pattern is judged to be the defect in the case where the Difference between the two exceeds the predetermined threshold. In this case, when the specific cell in the nth chip is the inspection target, the optical image preceding the specific cell becomes the reference image to be compared. Assuming, for example, that the second cell is the inspection target in cell A of FIG 10 is, the optical image of the first cell is the reference image.

For example, the width determination threshold registered as the line width defect is assigned in units of line width dimension differences (nm) and units of dimensional ratios (%). For example, the determination thresholds are assigned to the line width dimension difference of 16 nm and the dimensional ratio of 8% in two ways. When the dimension difference with the reference data is 20 nm while the pattern of the optical image data 204 , which becomes the inspection target, has a line width of 200 nm because the pattern is larger than both threshold values of the dimensional difference and the dimensional ratio, the pattern is registered as the defect.

The threshold value of the defect determination may be assigned separately for the case that the line width is larger than that of the reference data, and the case that the line width is smaller than that of the reference data. The threshold value can be assigned in both the case where the inter-pattern distance is not larger than that of the reference data, and the inter-pattern distance is smaller than that of the reference data. The threshold values of the hole diameter and the diameter dimension ratio may be assigned to the pattern having the hole shape. In this case, the threshold value for the sections in the X direction and Y direction of the hole may be assigned.

When compared with the cell comparison method, substantially all of the chips are set to the inspection area in the die-to-die comparison methods, and the inspection can regardless of the part in which the repetitive pattern exists. However, because the separation of the "dies" is greater than the distance between the cells that are compared with each other, the die-to-die comparison method is easily affected by the dimension distribution in the surface of the mask. That is, when the chips in the regions having the different dimensions are compared with each other, the patterns are compared to each other with dimensional advances (deviations), resulting in a problem that the defect to be detected can not be detected or the shape or Line width that does not need to be detected when the defect is detected. Therefore, in some cases, the result of the to-the-match method with the result of the cell comparison method is replaced to remove the influence of the dimension distribution in the surface of the mask from the acquired data. A specific technique is described below.

(Process of determining whether dimension distribution in the surrounding area of a defect detecting position falls within the predetermined range by comparing the dimension distribution in the vicinity of a defect detecting position with the dimension distribution of another region)

6 illustrates an example of the dimension difference map of the mask. In the example of 6 the line width is greater than the reference value in regions A and C. On the other hand, the line width is smaller than the reference value in regions B and D. 7 to 9 are views showing the dimensional distribution corresponding to the section taken along the line XX 'in FIG 6 illustrate. In the 7 to 9 For example, a horizontal axis has the same scale and an origin is positioned at the same position.

7 illustrates the dimensional distribution according to the map in 6 , 8th illustrates the dimensional distribution of the pattern measured by the dimension measuring circuit. In 8th For example, an absolute value of the dimensional difference in the area surrounded by the broken line is smaller than the absolute values of the dimensional differences in the remaining four points. Accordingly, at a glance the point in the broken line does not seem to be the defect. On the other hand, the absolute values of the dimensional differences in the area surrounded by an alternatively long and short dashed line are larger than the absolute value of the dimensional difference in the remaining two points. Accordingly, the dots in the alternating long and short dashed line appear to be the defect.

However, like out 7 As can be seen, the area of the area surrounded by the broken line (P2) is the region where the dimensional difference becomes negative. That is, the line width of the surrounding area is smaller than the reference value. On the other hand, because the dimensional difference in the part surrounded by the broken line is the positive value, the area surrounded by the broken line has a line width larger than the surrounding area. 9 illustrates the dimensional difference from the reference value of the line width in the measured part after the influence of the dimension distribution is removed. How out 9 That is, when the dimension difference in the part surrounded by the broken line (P2) becomes positive, the part surrounded by the broken line is unusually large compared to the surrounding area from the value of the positive dimensional difference, and therefore the part surrounded by the broken line should as the defect are detected.

On the other hand, how out 7 As can be seen, the surrounding area of the dot-dashed line (P1) in FIG 8th surrounded part of the region where the dimensional difference becomes positive. According to shows in 8th the dimensional difference of the part indicates the large positive value as a result of adding the trend of the line width distribution in the region. In 9 Although the dimensional difference of the part surrounded by the dot-dashed line (P1) has the positive value, the dimensional difference falls within the acceptable range and the point should not be detected as the defect.

In the die-to-die comparison method, the chips in regions A and B are in 6 compared to each other. Regions A and B have contradictory trends in line widths, as in 7 illustrated. Therefore, if the defect determination is made from the result of the dimension distribution, there is a possibility that the defect to be detected can not be detected, or a defect which does not need to be detected is detected as a defect.

In the region A where the line width is increased as a whole, the dimensional difference of the position where the line width is further increased will indicate the large value even if the dimensional difference of the position is practically acceptable. On the other hand, in the region B where the line width as a whole is reduced, the dimensional difference of the position where the line width is increased shows the small value even if the dimensional difference of the position is practically unacceptable. Therefore, when the regions A and B are compared with each other, the large dimension difference indicative region A is determined to be the defect the region B indicating the small dimension difference is determined not to be the defect.

In such cases, regions A and B are not compared but the cells in region A are compared or the cells in region B are compared. That is, the cell comparison method is used in place of the die-to-the-compare method. In the cell comparison method, because the regions having the same line width tendency are compared with each other, the comparison in the state of 9 are performed after the influence of the dimension distribution is removed. Because the defect that does not need to be detected is removed from the mask inspection result, the number of defects viewed by the operator is reduced to shorten the inspection time. Also, because the number of defects described in the defect information list is lowered, the production yield of the mask can be improved. In addition, the defect, which can hardly be detected due to the influence of the dimensional distribution, can be detected.

As to which of the results of the die-to-die comparison method and the result of the cell comparison method is used, this will be explained with reference to the flow chart in FIG 5 certainly.

In the case where the alignment measuring circuit 125 the pattern dimension parallel to the detection of the optical image in the mask 101 The last data in the data is measured by the alignment measurement circuit 125 measured dimensional difference data referenced when the defect is detected by the die-to-the-comparison method. The dimension distribution from the position where the defect is detected to the previous position where the dimensional difference is detected is compared with the dimension distribution in the chip and the dimension distribution between the chips (S16). As a result of the comparison, when the dimension distribution from the position where the defect is detected to the previous position where the dimensional difference is detected is determined to be within the predetermined range in S17 of FIG 5 falling, the result of the die-to-die comparison process, namely, the defect coordinate and the optical image and reference image, which are the basis of the defect determination, becomes as the mask inspection result 205 in the magnetic disk drive 109 saved.

In the case where the alignment measuring circuit 125 the pattern dimension parallel to the inspection of the comparison circuit 108 At the time when the defect is detected by the to-the-comparison method, the dimension of the pattern in which the comparison is already performed is measured, but becomes the dimension of the pattern in which the comparison is not performed is not measured. Therefore, in this case, the most recent data is output from the alignment measurement circuit 125 measured dimension difference data referenced.

On the other hand, the defect can be detected by the die-to-die comparison method, and the dimension distribution from the position where the defect is detected to the previous position where the dimensional difference is detected exceeds the predetermined range by comparing the dimension distribution with the Embodiment in the chip and the dimension distribution between the chips. In this case, the result of the die-to-die comparison method is not used in relation to the position, but the result of the parallel cell comparison method is used. At this point, whether the defect is detected as a result of the cell comparison method is not a problem. That is, the position determined to be a defect by the to-the-match method is not registered as the defect unless the position is determined to be defective by the cell comparison method.

However, in the case where a cell having a repetitive pattern that could be used as a reference does not exist near the cell that becomes the inspection target, the processing is performed only by the to-the-die comparison method. In this case, even if the dimensional distribution from the position where the defect is detected to the previous position where the dimensional difference is detected exceeds the predetermined range, by comparing the dimensional distribution with the dimension distribution in the chip and the dimension distribution between the chips, it is preferable to use the result of the die-to-die comparison method. That is, the coordinate of the detected defect and the optical image and reference image which are the basis of the defect determination are called the mask inspection result 205 on the magnetic disk drive 109 get saved.

By way of example, the dimensional distribution from the position where the defect is detected to the previous position where the dimensional difference is detected is compared with the dimensional distribution in the chip and the dimensional distribution between the chips. Alternatively, the dimension distribution from the position where the defect is detected to the previous position where the dimensional difference is detected is compared with the dimension distribution of the region separated by the chip pitch.

For example, the dimension distribution from the position where the defect is detected to the previous position where the dimensional difference is detected may be compared with (1) the dimension distribution in the strip containing the defect position or (2) the dimensional distribution of the Chip pattern, which is assumed from the strip, in which the dimensional difference is determined in advance, the strip containing the position determined as a defect. Whether that's out through the alignment measurement circuit 125 measured dimensional difference determined within the predetermined range falls, by comparing the dimensional distribution with (1) the dimension distribution or (2) the dimensional distribution is determined. When the dimensional distribution falls within the predetermined range, the coordinate of the defect detected by the to-the-thickness comparison method and the optical image and the reference image that are the basis of the defect determination may be used as the mask inspection result 205 on the magnetic disk drive 109 get saved. At this point, (1) the dimension distribution and (2) the dimension distribution are calculated by the map generating circuit 126 derived map derived. (1) the dimension distribution and (2) the dimension distribution can also be directly out of those by the alignment measurement circuit 125 derived dimension difference data are derived.

In the above modification, the defect is detected by the die-to-die comparison method, and the dimension distribution from the position where the defect is detected to the previous position where the dimensional difference is detected is determined by (1) the dimension distribution and (2 ) of the dimension distribution. When the dimensional distribution deviates from the position where the defect is detected to the previous position where the dimensional difference is detected from the predetermined range, the result of the to-the-comparison method is not used with respect to the position, but becomes Result of parallel cell comparison procedure used. In this case, whether the defect is detected as a result of the cell comparison method is not a problem. That is, even the position determined by the to-the-comparison method as the defect is not registered as the defect unless the position is determined as the defect by the cell comparison method. Alternatively, it may be determined whether the result of the to-the-match method is stored.

In the case that the cell which becomes the reference does not exist because the repetitive pattern does not exist near the cell which becomes the inspection target, the processing is performed only by the to-the-die comparison method. In this case, even if the dimension distribution deviates from the predetermined range from the position where the defect is detected to the previous position where the dimension difference is detected, by comparing the dimension distribution with (1) the dimension distribution or (2) Dimensional distribution, the result of the die-to-die comparison method is used. When the mask as a whole is viewed, the addition of the defect that does not need to be detected is reduced to the mask inspection result, and the defect that is hardly detected due to the influence of the dimension distribution can be detected.

In the second embodiment, the control computer 110 in 1 determine whether the result of the die-to-die method is appropriately used by comparing the dimension distribution in the chip and the dimension distribution between the chips having the dimension distribution from the position where the defect is detected to the previous position where the dimension difference is determined. Alternatively, the control computer 110 determine whether the result of the die-to-die method is suitably used by comparing the dimension distribution from the position where the defect is detected to the previous position where the dimensional difference is determined by (1) the dimension distribution or (2) the dimension distribution. The control computer 110 can determine if the result of the cell comparison process exists. In the case where the result of the die-to-compare method does not exist as the result of the cell comparison process but only the result of the die-to-compare method, the result of the die-to-die comparison process becomes the mask inspection result 205 in the magnetic disk drive 109 stored independently of the comparison result.

When passing through the alignment measuring circuit 125 If the detected dimensional difference has a large degree, the position can be registered as the defect even if the position is not considered the defect in the comparison circuit 108 is determined. Therefore, in the second embodiment, the defect determination can be made as follows.

For example, it is assumed that the threshold value by which the comparison circuit 108 detects the line width defect as a line width dimension difference of 16 nm and an aspect ratio of 8% is set. The threshold value of the defect determination for the measurement result of the alignment measurement circuit 125 is compared to the threshold value by which the comparison circuit 108 determines the line width defect, slightly relaxed, the line width dimension difference is set to 20 nm, and the aspect ratio is set to 10%. With regard to the predetermined range which becomes the criterion at which either the result of the die-to-die comparison method or the result of the cell comparison method is used, the line width dimension difference is set to 12 nm or more and the aspect ratio becomes 6 % or more is set.

The result of the die-to-the-match method is used when passing through the alignment measurement circuit 125 determined dimensional difference is smaller than 12 nm, while the dimensional ratio is smaller than 6%. On the other hand, the result of the cell comparison method is used when passing through the alignment measurement circuit 125 The measured dimensional difference is greater than or equal to 12 nm and less than 20 nm, while the dimensional ratio is greater than or equal to 6% and less than 10%. The position is registered as the defect when passing through the alignment measurement circuit 125 The measured dimensional difference is greater than or equal to 20 nm while the dimensional ratio is greater than or equal to 10%.

The predetermined range which becomes the criterion to which either the result of the to-the-match method or the result of the cell comparison method is applied is set in each mask which becomes the inspection target. At this point, the predetermined range is set to the range that does not exceed the threshold in the case where the position is determined as the defect, from the measurement value of the alignment measuring circuit 125 , The adjustment method is similar to the threshold adjustment method similar to the comparison circuit 108 , That is, the predetermined range may be assigned individually for the case where the line width is larger than the reference data and the case where the line width is smaller than the reference data and the predetermined range may be assigned for the case not the line width, but the inter-pattern distance is larger than the reference data and the case where the inter-pattern distance is smaller than the reference data. In addition, the predetermined range of the hole diameter or the dimension ratio may be assigned to the diameter of the pattern having the hole shape. In this case, the predetermined range may be assigned to both the X-direction and Y-direction portions of the hole.

In the second embodiment, the data generated by the card generating circuit 126 generated map used to the pattern in the mask 101 to transfer to the wafer. For example, if the exposure device containing the pattern in the mask 101 transfers to the wafer, which can input irradiation energy (dose) as the card, which is passed through the card generating circuit 126 generated map on the exposure device and converted into the map of the irradiation energy, which allows the line width is transmitted homogeneously to the wafer. For example, in the position where the dimensional difference in the mask 101 becomes negative, namely the position at which the line width is thinned, the irradiation energy is adjusted so that the pattern transmitted to the wafer is thickened. On the other hand, in the position where the dimensional difference in the mask 101 becomes positive, namely, the position where the line width is thickened, the irradiation energy is adjusted so that the pattern transmitted to the wafer is thinned. Therefore, the line width of the pattern transferred to the wafer is homogenized even in the mask in which the pattern has the dimension distribution.

According to the present invention, an inspection apparatus includes an optical image detection unit that virtually divides a sample in a plurality of stripe-shaped stripes along a predetermined direction to detect an optical image of the sample in each of the stripes, a reference image generation unit that performs filtering based on Performs design data of the chip pattern formed on the sample to generate a reference image corresponding to the optical image, a first comparator which outputs the chip pattern of the optical image output from the optical image detecting unit with the chip pattern of the optical pattern comparing a reference image generated from the reference image generation unit by a die-to-database comparison method, a second comparator comparing repetitive pattern regions in the chip pattern of the optical image output from the optical image detection unit using a cell method, a dimension difference / dimension ratio Erf A mapping unit that obtains a dimensional difference or / and a dimensional ratio between a pattern of the optical image and a pattern of the reference image, compared with the pattern of the optical image by the to-database method, a dimension distribution acquiring unit having a dimension distribution of the plurality of chip Patterns of the dimensional difference or / and the dimension ratio determined, which are output from the dimension difference / dimension ratio detection unit, and a controller that stores a result of the first comparator, when compared to a designated as a defect place by the comparison of the first Comparator, a dimension distribution from the place to the previous place, where the dimensional distribution or / and the dimension ratio is determined by the dimensional difference / dimension ratio detection unit falls within a predetermined range, by comparing the dimension distribution with the Dimension distribution in the strip, which contains the space determined as a defect, or one Dimensional distribution of the chip pattern, which is guessed from the strip in which the dimensional difference is determined in advance, of the strip concerned, and stores a result of the second comparator instead of the result of the first comparator, if the dimension distribution from the space determined to be defect to previous space where the dimensional difference or / and the dimension ratio is detected by the dimensional difference / dimension ratio detection unit exceeds the predetermined range by comparing the dimension distribution with the dimension distribution in the strip including the space determined as a defect or the dimensional distribution of the chip Pattern estimated from the strip in which the dimensional difference is detected in advance of the strip concerned.

The controller controls the result of the first comparator regardless of the dimension distribution from the space determined to be defect to the previous place where the dimensional difference or / and the dimension ratio is determined by the dimension difference / dimension ratio detection unit, if the result of the second comparator does not exist, because the repetitive pattern area does not exist in the space determined by the comparison of the first comparator as a defect.

Further, according to the present invention, an inspection apparatus includes an optical image detecting unit that virtually divides a sample in a plurality of stripe-shaped stripes along a predetermined direction to obtain an optical image of the sample in each of the stripes, a first comparator defining the chip patterns comparing the optical image output from the optical image detecting unit with a die-to-die method, comparing a second comparator comparing repetitive pattern areas in the chip pattern of the optical image output from the optical image detecting unit with a cell method, a dimension difference / dimension ratio detecting unit, determining a dimensional difference or / and a dimensional ratio between a pattern of the optical image and a pattern of the reference image as compared with the pattern of the optical image by the die-to-die method, a dimension distribution detecting unit having a dimension distribution of the plurality of chip patterns from the dimensional difference or / and the dimension ratio output from the dimension difference / dimension ratio detection unit, and a controller storing a result of the first comparator with respect to a detected by the comparison of the first comparator as a defect Space, a dimension distribution from the place to a previous place where the dimensional difference or / and the dimension ratio is determined by the dimensional difference / dimension ratio detection unit falls within a predetermined range, by comparing the dimension distribution with a dimension distribution in the strip containing the space , which is determined as a defect, or a dimensional distribution of the chip pattern estimated from the strip in which the dimensional difference is detected in advance of the strip concerned, and stores a result of the second comparator instead of the result of the first compara When the dimensional distribution from the space determined to be a defect to the previous place where the dimensional difference or / and the dimensional ratio is determined by the dimensional difference / dimension ratio detection unit exceeds the predetermined range, by comparing the dimension distribution with the dimension distribution in the strip contains the space determined as a defect, or the dimensional distribution of the chip pattern estimated from the strip in which the dimensional difference is detected in advance of the strip concerned.

The controller stores the result of the first comparator regardless of the dimension distribution from the space designated as a defect to the previous place where the dimensional difference or / and the dimension ratio is determined by the dimension difference / dimension ratio detection unit, if the result of the second comparator does not exist, because the repetitive pattern area does not exist in the place determined by the comparison of the first comparator as a defect.

The present invention is not limited to the described embodiments and can be implemented in various ways without departing from the spirit of the invention.

The above description of the present embodiment has no specified device structures, control methods, etc., which are not essential to the description of the invention, since any suitable apparatus construction, control method, etc. can be used to implement the invention. Further, the scope of this invention encompasses all supporting devices employing the elements of the inventions and variations thereof which may be devised by those skilled in the art.

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• JP 2013-055500 [0001]
• JP 4236825 [0005]
• JP 444768 [0007]
• JP 3824524 [0071]

Claims (13)

Inspection procedure, comprising: virtually dividing a sample in which a plurality of chip patterns are formed into a plurality of strip-shaped strips along a predetermined direction to detect an optical image of the chip pattern in each of the strips; Performing filtering based on design data of the chip pattern to produce a reference image corresponding to the optical image; Comparing the chip pattern using a die-to-database comparison method and comparing a repetitive pattern portion in the chip pattern using a cell method; Determining a dimensional difference or / and a dimensional ratio between a pattern of the optical image and a pattern of the reference image compared with the pattern of the optical image by the to-database comparison method, and determining a dimension distribution of the plurality of chip patterns the dimensional difference or / and the dimension ratio, wherein, with respect to a position determined to be a defect by the comparison of the die-to-database comparison method, a result of the die-to-database comparison method is stored when a dimension distribution from the position to a previous position where the Dimensional difference or / and the dimensional ratio is determined falls within a predetermined range by comparing the dimensional distribution with a dimension distribution in the strip containing the position determined as a defect or a dimension distribution of the chip pattern adopted from the strip in which the dimension difference is determined in advance of the strip concerned, and a result of the cell method is stored instead of the result of the die-to-database comparison method, when the dimension distribution from the position determined as a defect to the previous position where the dimension difference or / and the dimensional ratio is determined t exceeds the predetermined range by comparing the dimension distribution with the dimension distribution in the strip containing the defect-determined position or the dimension distribution of the chip pattern adopted from the strip in which the dimensional difference of the strip concerned is determined in advance , Inspection procedure, comprising: Detecting an optical image of a sample in which a plurality of chip patterns are formed, performing filtering based on design data of the chip pattern to generate a reference image corresponding to the optical image; Comparing the chip pattern using a die-to-database comparison method and comparing a repetitive pattern portion in the chip pattern using a cell method; Determining a dimensional difference or / and a dimensional ratio between a pattern of the optical image and a pattern of the reference image that is compared with the pattern of the optical image by the to-database comparison method, and Determining a dimension distribution of the plurality of chip patterns from the dimensional difference or / and the dimension ratio, wherein, with respect to a position determined to be a defect by the comparison of the die-to-database comparison method, a result of the die-to-database comparison method is stored when a dimension distribution from the position to a previous position where the Dimension difference or / and the dimensional ratio is determined falls within a predetermined range, by comparing the dimensional distribution with a dimensional distribution in a chip or a dimension distribution between chips, and a result of the cell method is stored instead of the result of the die-to-database comparison method; when the dimension distribution from the position determined as a defect to the previous position where the dimensional difference or / and the dimension ratio is determined exceeds the predetermined range, by comparing the dimension distribution with the dimension distribution in the chip, or the dimension distribution between the chi ps. An inspection method according to claim 1 or 2, wherein the result of the die-to-database method is stored regardless of the dimension distribution from the position determined as a defect to the previous position where the dimension difference or / and the dimension ratio is determined the result of the cell comparison does not exist because the repetitive pattern part does not exist in the position determined by the comparison of the die-to-die method as a defect. An inspection method according to any one of claims 1 to 3, wherein the dimensional difference is a difference in the line width between the pattern of the optical image and the pattern of the reference image, or a difference in the distance between the patterns of the optical image and a distance between the patterns of the reference image , An inspection method according to any one of claims 1 to 4, wherein the dimension ratio is a line width ratio of the pattern of the optical image and the pattern of the reference image, or a ratio of a distance between the patterns of the optical image and a distance between the patterns of the reference image. An inspection method comprising: virtual dividing a sample in which a plurality of chip patterns are formed into a plurality of stripe-shaped strips along a predetermined direction to obtain an optical image of the chip pattern in each of the strips; Comparing the chip pattern by a die-to-die method and comparing a repetitive pattern part in the chip pattern by a cell method; Determining a dimensional difference or / and a dimensional ratio between a pattern of the optical image and a pattern of the reference image as compared with the pattern of the optical image by the die-to-die methods; and determining a dimensional distribution of the plurality of chip patterns from the dimensional difference or / and the dimension ratio, wherein a result of the die-in relation to a position which is determined by the comparison of the die-to-die method as a defect The method is stored when a dimension distribution from the position to a previous position where the dimensional difference or / and the dimension ratio is detected falls within a predetermined range by comparing the dimensional distribution with a dimension distribution in the strip that determined the defect Position, or a dimension distribution of the chip pattern, which is assumed from the strip in which the dimensional difference is determined in advance of the strip concerned, and a result of the cell method is stored instead of the result of the to-the-the-method, if the Dimension distribution from the position intended as a defect to before 4, where the dimensional difference or / and the dimension ratio is determined exceeds the predetermined range by comparing the dimension distribution with the dimension distribution in the strip containing the position determined as the defect or the dimension distribution of the chip pattern is assumed from the strip, wherein the dimension difference is determined in advance to the affected strip. Inspection procedure, comprising: Determining an optical image in which a plurality of chip patterns are formed; Comparing the chip pattern by a die-to-die method and comparing a repetitive pattern part in the chip pattern by a cell method; Determining a dimensional difference or / and a dimensional ratio between a pattern of the optical image and a pattern of the reference image as compared with the pattern of the optical image by the die-to-die methods; and Determining a dimension distribution of the plurality of chip patterns from the dimensional difference or / and the dimension ratio, wherein, with respect to a position determined by the comparison of the die-to-die method as a defect, a result of the die-to-die method is stored when a dimension distribution from the position to a previous position in which the dimensional difference or / and the dimensional ratio is detected, falls within a predetermined range, by comparing the dimensional distribution with a dimension distribution in a chip or a dimension distribution between chips, and a result of the cell method is stored instead of the result of the the-to-die method when the dimension distribution from the position determined as a defect to the previous position where the dimensional difference or / and the dimension ratio is determined exceeds the predetermined range, by comparing the dimension distribution with the dimension distribution of the dimension distribution in the chip or the dimension distribution between chips , An inspection method according to claim 6 or 7, wherein the result of the die-to-die process is stored independently of the dimension distribution from the position determined as a defect to the previous position where the dimension difference or / and the dimension ratio is determined the result of the cell method does not exist because the repetitive pattern part does not exist in the position determined by the comparison of the die-to-die method as a defect. The inspection method according to any one of claims 6 to 8, wherein the dimensional difference is a difference in the line width between the patterns of the optical images or a difference in the distance between the patterns of the optical images. An inspection method according to any one of claims 6 to 9, wherein the dimensional ratio is a line width ratio of the patterns of the optical images or a ratio of a distance between the patterns of the optical image. An inspection apparatus comprising: an optical image detecting unit (A) that virtually divides a sample into a plurality of stripe-shaped stripes along a predetermined direction to detect an optical image of the sample in each of the stripes; a reference image generating unit (10); 112 ) performing filtering based on design data of the chip pattern formed on the sample to produce a reference image corresponding to the optical image, a first comparator (Fig. 108a ) which displays the chip pattern of the optical image output from the optical image detecting unit (A) with the chip pattern of the reference image generating unit (Fig. 112 ) compares the output reference image with a database-to-database method, a second comparator ( 108b ) comparing repetitive pattern parts in the chip pattern of the optical image output from the optical image detecting unit (A) using a cell method, a dimension difference / dimension ratio detecting unit (FIG. 125 ), which determines a dimensional difference or / and a dimensional ratio between a pattern of the optical image and a pattern of the reference image, as compared with the pattern of the optical image by the die-to-database method, a dimension distribution detecting unit (FIG. 126 ) which determines a dimension distribution of the plurality of chip patterns from the dimensional difference or / and the dimensional ratio, which is calculated from the dimensional difference / dimension ratio detection unit (FIG. 125 ) and a controller ( 110 ), which is a result of the first comparator ( 108a ) when, in relation to a location determined by the comparison of the first comparator ( 108a ) is determined as a defect, a dimension distribution from the place to a previous place, the dimension difference or / and the dimension ratio by the dimension difference / dimension ratio detection unit (FIG. 125 ), falls within a predetermined range, by comparing the dimension distribution with a dimension distribution in the strip containing the space determined as a defect or a dimension distribution of the chip pattern suspected from the strip in which the dimension difference is in advance affected strip, and a result of the second comparator ( 108b ) instead of the result of the first comparator ( 108a ) stores, when the dimension distribution of the space designated as a defect to the previous place where the dimensional difference or / and the dimensional ratio by the dimension difference / dimension ratio detecting unit (FIG. 125 ) exceeds the predetermined range, by comparing the dimensional distribution with the dimension distribution in the strip containing the defect-specific space or the dimension ratio of the chip pattern suspected from the strip in which the dimension difference is detected in advance of the stripe concerned. An inspection apparatus comprising: an optical image detecting unit (A) that virtually divides a sample into a plurality of stripe-shaped stripes along a predetermined direction to detect an optical image of the sample in each of the stripes; a first comparator (A); 108a ) comparing the chip patterns of the optical image output from the optical image detecting unit (A) by a die-to-die method, a second comparator ( 108b ) comparing repetitive pattern parts in the chip pattern of the optical image output from the optical image detecting unit (A) by a cell method, a dimension difference / dimension ratio detecting unit (FIG. 125 ), which determines a dimensional difference or / and a dimensional ratio between a pattern of the optical image and a pattern of the reference image compared to the pattern of the optical image by the die-to-die method, a dimension distribution detecting unit (FIG. 126 ) which determines a dimension distribution of the plurality of chip patterns from the dimensional difference or / and the dimensional ratio, which is calculated from the dimensional difference / dimension ratio detection unit (FIG. 125 ) and a controller ( 110 ), which is a result of the first comparator ( 108a ) when, in relation to a location determined by the comparison of the first comparator ( 108a ) is determined as a defect, a dimension distribution from the place to a previous place, the dimension difference or / and the dimension ratio by the dimension difference / dimension ratio detection unit (FIG. 125 ), falls within a predetermined range, by comparing the dimension distribution with a dimension distribution in the strip containing the space determined as a defect or a dimension distribution of the chip pattern suspected from the strip in which the dimension difference is in advance affected strip, and a result of the second comparator ( 108b ) instead of the result of the first comparator ( 108a ) stores, when the dimension distribution of the space designated as a defect to the previous place where the dimensional difference or / and the dimensional ratio by the dimension difference / dimension ratio detecting unit (FIG. 125 ) exceeds the predetermined range, by comparing the dimensional distribution with the dimension distribution in the strip containing the defect-determined space or the dimension distribution of the chip pattern suspected from the strip in which the dimensional difference is detected in advance of the stripe concerned. Inspection device according to one of claims 11 or 12, wherein the controller ( 110 ) the result of the first comparator ( 108a ) regardless of the dimension distribution from the place designated as a defect to the previous place where the dimension difference or / and the dimension ratio by the dimension difference / dimension ratio detection unit (FIG. 125 ) is determined when the result of the second comparator ( 108b ) does not exist because the repetitive pattern part does not exist in the place that is a defect determined by comparing the first comparator ( 108a ).

Images